Monday, October 13, 2008

NPS revisited

Here we go again. There is another article about NPS - also critical. The article was published in the October, 2008 issue of Quirk's Marketing Research Review under the title "Customer Loyalty 2.0" NPS (Net Promoter Score) is a simple methodology for indicating a degree of loyalty that, according to the developer of NPS, is a good predictor of company growth by asking one question; namely "how likely is it that you would recommnd this company to your friend or colleague."

This methodology has come under some very severe criticism by both the academic community and the consultants who specialize on the measurement of customer loyalty. Some of the researchers found that customer satisfaction is just as good at predicting growth. Other researchers have found that measures such as likelihood to repurchase are also comparable to NPS for predicting growth.

The author of the article makes the statement that there is no published empirical evidence supporting the superiority of NPS over the other conventional loyalty metrics. The basis for this statement is that recent scientific, peer-reviewed studies do not appear to support the claims of NPS. The author makes two very strong requests of the developers of NPS; namely,
1. Present their analysis to back up their claims.
2. Refute the current scientific research that brings their methodological rigor into question.

It seems that most researchers have found that customer satisfaction is consistently correlated with growth. I have found (and published) results that show that high levels of customer satisfaction yield better financial performance for companies than those with low levels of customer satisfaction.

The author and researcher then conducted a statistical test using 1,000 respondents in the United States (18 years and older). [I assume it was a valid statistical sample]. The test asked them the following four questions:
1. How satisfied are you with Company ABC
2. How likely are you to recommend Comapny ABC
3. How likely are you to continue to purchasing the same product and/or service from Company ABC
4. If you were selecting a company for the first time, how likely is it that you would choose Company ABC.

The result was that all four questions appeared to measure one underlying construct, customer loyalty. There was a correlation of .87 among the four questions. Needless to say the statistics appear to indicate the four questions could be considered interchangeable as a direct measure of loyalty and using the NPS logic each of the four could also be used as a good predictor of company growth.

I think the author of the article makes a great point when he points out that a single item measure is less reliable than a multiple item meaure. He uses the example of using a single question to determine the SAT score. If that were the case then why bother with the other questions. In fact, the author averages the scores of all four questionsshown above and suggests that the average of all four measures is a more precise measure of loyalty than any of the four questions used alone.

One point that the author makes is that the customers who score 7 or 8 may be potential disloyal customers and are currently overlooked by NPS. He cites a study of wireless service providers where he found that 31 percent of those who scored 7 or 8 were likely to switch to a different wireless service provider even though they were not identified as a detractor.

In conclusion, the author suggests that companies should use a variety of loyalty questions so that those customers who may be likely to switch can be identified. To quote the author "how well we are able to predict business performance masures depends on the match between the business metric and the loyalty questions."

The bottom line is that the people who are selling NPS, while it is being touted as the best measure, need to publish more data to support their claims. There are some serious claims being made that strongly suggests that NPS is not all it is advertised to be. Time will tell.

No comments:


web visitor stats
OptiPlex 755 Desktops